• Measurements: First and Second Generation Apple AirPort Express

    Seven years ago Stereophile's John Atkinson published objective measurements of Apple's AirPort Express (Link ). Since then Apple changed critical internal parts increasing jitter levels and more recently completed a redesign of the entire AirPort Express. The time has come to put both new and old Express units on the analyzer and publish updated measurements.











    Products Measured

    First Generation AirPort Express model number A1264, serial number 6F149A8Y2UF running firmware version 7.6.1




    Second Generation AirPort Express model number A1392, serial number C86HT2TVDV2R running firmware version 7.6.2




    Testing was conducted by an independent lab using Mac OS X 10.6.8, iTunes 10.6.3, optical digital output, and analog output from the AirPort Express into an Audio Precision analyzer. The first generation AirPort Express used in these tests was purchased after the release of the second generation unit. This was the last production version of the first generation model. Readers should know that I am not capable of performing the tests or analyzing the results without an extraordinary amount of help from the testing lab. I've attempted to explain some of the results in general terms but many of the results are far too complex for simple explanations. As such I have used wording directly from the testing lab in many instances.






    Executive Summary

    First Generation AirPort Express - WAV, AIFF, Apple Lossless, MP3, and AAC files at 16 bit / 44.1 kHz are bit perfect (without data loss) from iTunes to the AirPort and through the AirPort's optical output. All other sample rates including 48 kHz, 88.2 kHz, 96 kHz, 176.4 kHz, and 192 kHz are converted to 44.1 kHz. All 24 bit content is truncated to 16 bits. The phase remains correct even with bit truncation from 24 bits to 16 bits. Both 16 bit DC and Walking Zero tests were bit true as well. 24 bit Walking Zero was truncated to 16 bit.

    Playback of a 997 Hz test tone measured at the digital output of the AirPort Express displayed a bit higher noise at 48 kHz. Playback of the same test tone at 176.4 kHz and 192 kHz displays severe distortion as the sample rate converter is overdriven.

    The AirPort Express always sent out the same status bit and Copyright bit.

    Warning: The following jitter numbers should not be used as the only indicator of the AirPort Express' performance. These numbers are but one indication of sound quality. The average jitter from 700 Hz to 100 kHz is about 474 picoseconds on the digital optical output. Peak jitter from 50 Hz to 100 kHz is about 1.52 nanoseconds on the digital optical output.



    Second Generation AirPort Express - WAV, AIFF, Apple Lossless, MP3, and AAC files at 16 bit / 44.1 kHz are bit perfect (without data loss) from iTunes to the AirPort and through the AirPort's optical output. All other sample rates including 48 kHz, 88.2 kHz, 96 kHz, 176.4 kHz, and 192 kHz are converted to 44.1 kHz. All 24 bit content is truncated to 16 bits. The phase remains correct even with bit truncation from 24 bits to 16 bits. Both 16 bit DC and Walking Zero tests were bit true as well. 24 bit Walking Zero was truncated to 16 bit.

    Playback of a 997 Hz test tone measured at the digital output of the AirPort Express displayed a bit higher noise at 48 kHz. Playback of the same test tone at 176.4 kHz and 192 kHz displays severe distortion as the sample rate converter is overdriven.

    The AirPort Express always sent out the same status bit and Copyright bit.

    Warning: The following jitter numbers should not be used as the only indicator of the AirPort Express' performance. These numbers are but one indication of sound quality. The average jitter from 700 Hz to 100 kHz is about 280 picoseconds on the digital optical output. Peak jitter from 50 Hz to 100 kHz is about 1.25 nanoseconds on the digital optical output.



    Comparison

    The newer AirPort Express, as previously mentioned, performs nearly identical to the first generation units with jitter being the major difference. The second generation AirPort Express shows a reduction of 41% (average) and 18% (peak) jitter on the digital optical output. Comparing jitter and THD on the analog output of the first and second generation AirPorts it's clear there has been a major change between the units. Jitter and THD from the second generation AE has increased dramatically. This increase has the largest impact on over all performance even for audiophiles using the digital output only. According to the independent testing lab, "In the digital out of the bi-phase signal, there is the bit clock embedded. And either from this bit clock the master clock is generated, or the bit clock is divided by the master clock. From this bit clock or / and master clock, the final analog signal is converted. The output of the analog signal is a sort of audio part image of the bit clock / master clock. When measuring the analog out of the AE, it shows the audio band related behavior of the bit clock, that is simultaneously sent via digital out to an outboard DAC. Even good outboard DACs will have lower jitter in the final converted analog signal, the sonic imprint of the remaining jitter has a strong correlation of the analog jitter of the internal converted analog signal. In the bi-phase signal (digital out) there are many more other frequencies than jitter numbers, when measuring the jitter on the digital out, but only those frequencies, that will convert into the audio band, will have a sonic impact into the audio band. So with the analog jitter measurement, you are measuring only that parts of jitter, that matter in the audio band."

    It's critical to understand the analog output of the internal AirPort Express DAC shows a sonic fingerprint of what is inside the digital out that will find its way to a external DAC. Examples of how the first and second generation units differ and why the first generation AirPort Express is sonically a better performer.


    An example of jitter measurements NOT real numbers.
    On the digital out of the Second Generation AirPort Express there is 1 nanosecond low frequency jitter (20 Hz to 200 Hz) and 1 nanosecond higher frequency jitter (over 50 kHz). A measurement will show in total 1.4 nanosecond jitter (square root out of 1 +1). When converting into analog the 50 kHz jitter will be suppressed by the receiver (every receiver has a low pass filter), but the 20 Hz jitter not and will still be 1 nanosecond at the analog out.

    An example of jitter measurements NOT real numbers.
    On the digital out of the First Generation AirPort Express there is 10 picosecond low frequency jitter and about 2 nanoseconds higher frequency jitter. A measurement at the digital out will tell you 2 nanoseconds jitter and appear worse to the layperson. When converted this signal into the audio band, only the 10 ps bass jitter will remain and as a result will sound much better even though the digital out itself looks worse.










    Details


    First Generation Apple AirPort Express Measurements


    Sample Rate Conversion (SRC) with a 997 Hz test tone measured at the digital optical output.

    1 FS and 2 FS Sample Rates: 44.1 kHz (blue), 48 kHz (red), 88.2 kHz (cyan), 96 kHz (magenta)
    At 16 Bit, all 1 FS and 2 FS show similar performance, only 48 kHz shows a bit higher noise.

    16 bit input and output


    24 bit input - 16 bit output



    4 FS Sample Rates: 176.4 kHz (cyan), 192 KHz (magenta), 44.1 kHz (blue), all at 24 Bit input – 16 Bit output
    176.4 kHz and 192 KHz are "too much" for the SRC. Besides 176.4 kHz and 192 kHz, where the sample rate converter was "overdriven" and does show severe distortion. Every other sample rate, 48 kHz, 88.2 kHz, 96 kHz, is sample rate converted to 44.1 kHz with approximately 16 Bit Performance and Resolution. Only 48 kHz shows a bit more degradation.





    Sample Rate Conversion (SRC) test with a synchronous test signal.

    The data output from iTunes to Airport Express is truncated down to 16 Bit. Every distortion from the 24 Bit SRC of the OS is hidden below the 16 Bit resolution (except for 176.4 kHz and 192 kHz, where the overload of the SRC can be seen above 16 Bit resolution). In order to have a better view of what is going on under the hood, we can use the 16 Bit synchronous J-Test Signal, because this signal has real "digital black bins" between the frequencies, that goes gown to infinity.

    When using the synchronous test signal we can see distortions that would be otherwise hidden below the 16 Bit resolution. Here we see the real distortions of the SRC that were otherwise buried in noise with the 997 asynchronous test tone.

    44k1 J-Test Signal (without SRC)



    48k J-Test Signal (with SRC to 44k1)








    Volume Control Testing

    16 Bit 997 Hz test tone with – 3 dBFS @ 0 dBr (blue) and – 10 dBr (red). Looks like the 24 Bit volume control of iTunes truncated down to 16 Bit with tiny distortions.



    16 Bit J-Test (synchronous with SR) @ 0 dBr (blue) and – 10 dBr (red). Modulation of the FS/192 LSB Signal with the FS/4 Pilot Tone due to level reduction.




    Status Bit Testing

    The unit always sends out the same status bit information and Copyright Bit. Channel Num B is not indicated as right channel.







    Jitter Testing - Digital Optical Output

    The Average Jitter (Top) with bandwidth from 700 Hz to 100 kHz is about 474 ps, and the Peak Jitter (Bottom), with bandwidth from 50 Hz to 100 kHz is about 1.52 ns.











    Jitter Testing - Analog Output

    44.1 kHz is blue and 48 kHz is red. Providing a single jitter number in this section is a disservice and isn't indicative of device performance. A major difference displayed in these two graphs is the appearance of SRC distortions in the 48 kHz red graph. Overall jitter is fairly low.












    THD Testing - Analog Output

    44.1 kHz is blue and 48 kHz is red. The Total Harmonic Distortion character doesn't have as smooth sounding of roll off as the Second Generation AirPort Express. There is no sign for Jitter in the analog output. It's presumed there is an internal part that is responsible for this low jitter output of the first generation AE (unlike the second generation AE).




















    Second Generation Apple AirPort Express Measurements


    Sample Rate Conversion (SRC) with a 997 Hz test tone measured at the digital optical output.

    1 FS and 2 FS Sample Rates: 44.1 kHz (blue), 48 kHz (red), 88.2 kHz (cyan), 96 kHz (magenta)
    At 16 Bit, all 1 FS and 2 FS show similar performance, only 48 kHz shows a bit higher noise.


    16 bit input and output


    24 bit input - 16 bit output



    4 FS Sample Rates: 176.4 kHz (cyan), 192 KHz (magenta), 44.1 kHz (blue), all at 24 Bit input – 16 Bit output
    176.4 kHz and 192 KHz are "too much" for the SRC. Besides 176.4 kHz and 192 kHz, where the sample rate converter was "overdriven" and does show severe distortion. Every other sample rate, 48 kHz, 88.2 kHz, 96 kHz, is sample rate converted to 44.1 kHz with approximately 16 Bit Performance and Resolution. Only 48 kHz shows a bit more degradation.





    Sample Rate Conversion (SRC) test with a synchronous test signal.

    The data output from iTunes to Airport Express is truncated down to 16 Bit. Every distortion from the 24 Bit SRC of the OS is hidden below the 16 Bit resolution (except for 176.4 kHz and 192 kHz, where the overload of the SRC can be seen above 16 Bit resolution). In order to have a better view of what is going on under the hood, we can use the 16 Bit synchronous J-Test Signal, because this signal has real "digital black bins" between the frequencies, that goes gown to infinity.

    When using the synchronous test signal we can see distortions that would be otherwise hidden below the 16 Bit resolution. Here we see the real distortions of the SRC that were otherwise buried in noise with the 997 asynchronous test tone.

    44k1 J-Test Signal (without SRC)



    48k J-Test Signal (with SRC to 44k1)








    Volume Control Testing

    16 Bit 997 Hz test tone with – 3 dBFS @ 0 dBr (blue) and – 10 dBr (red). Looks like the 24 Bit volume control of iTunes truncated down to 16 Bit with tiny distortions.



    16 Bit J-Test (synchronous with SR) @ 0 dBr (blue) and – 10 dBr (red). Modulation of the FS/192 LSB Signal with the FS/4 Pilot Tone due to level reduction.





    Status Bit Testing

    The unit always sends out the same status bit information and Copyright Bit.









    Jitter Testing - Digital Optical Output

    The Average Jitter (Top) with bandwidth from 700 Hz to 100 kHz is about 280 ps, and the Peak Jitter (Bottom), with bandwidth from 50 Hz to 100 kHz is about 1.25 ns.








    Jitter Testing - Analog Output

    44.1 kHz is blue and 48 kHz is red. Providing a single jitter number in this section is a disservice and isn't indicative of device performance. Overall jitter is much higher from the second generation AirPort Express than the original device.

    To “analyze” this graph: 11.025 kHz (12000 kHz) is the Pilot frequency and right and left of this center, is coming first the modulation of the bass. With this FFT resolution, approximately – 127 dBr is 16 Bit resolution (see the graph, perfect 16 Bit Jitter Signal), so the bass area, that is +/- 100 Hz beside the pilot tone, the graph is at about -112 dB, and this is 15 dB above 16 Bit threshold, or in other words the bass is only 13.5 Bit resolution in his timing.










    THD Testing - Analog Output

    44.1 kHz is blue and 48 kHz is red. The Total Harmonic Distortion has a natural roll off. The Jitter Performance is the worst part of this device and dominates everything. This is really very bad. It's presumed there is absolutely nothing included to reduce jitter in the second generation AirPort Express.










    Comments 33 Comments
    1. jhwalker's Avatar
      jhwalker -
      So am I reading this properly: the new (second?) generation Apple Express has lower jitter over the digital (Toslink) out, but very bad jitter if you're using analog out? So if I'm using the AE purely as a streamer into a remote DAC, the jitter (though still high) will not be nearly so bad as if I were using the analog out?

      Just want to make sure I'm reading this properly.
    1. Old Listener's Avatar
      Old Listener -
      Chris,

      Thanks for posting this article. I'm not in the market for an old or new Airport Express but I like to see measurements for audio gear and learn from what I see.

      Bill
    1. mayhem13's Avatar
      mayhem13 -
      Very Nice Chris!....From an avid Apple streamer....thank you!
    1. MikeJazz's Avatar
      MikeJazz -
      This is an opportunity for an after-market service....an "Audiophile" Airport Express, working as a pure streamer to a external DAC would be very nice...
      Altough it would have to be done underground...I guess...
    1. TimU's Avatar
      TimU -
      Quote Originally Posted by jhwalker View Post
      So am I reading this properly: the new (second?) generation Apple Express has lower jitter over the digital (Toslink) out, but very bad jitter if you're using analog out? So if I'm using the AE purely as a streamer into a remote DAC, the jitter (though still high) will not be nearly so bad as if I were using the analog out?

      Just want to make sure I'm reading this properly.
      I am also confused. It would be helpful if the conclusions were stated more clearly.
    1. JR_Audio's Avatar
      JR_Audio -
      Fingerprint

      The analog output, is time wise build by the same master clock, that drives the bit cells at the digital out. So the jitter performance of the analog out has some good correlations to the audio band part of the jitter at the digital out.

      Taking this into account, even your outboard DAC has better jitter suppression than the internal DAC of the AE, the sonic fingerprint of the jitter will find also its way to the analog out of the outboard DAC.

      To make a long story short, yes, the AE First Gen will have a lower jitter and better sound with your outboard DAC, than the AE Second Gen. An outboard DAC can only suppress incoming jitter, but can never remove.

      Juergen
    1. Akapod's Avatar
      Akapod -
      I think I get it.

      "The second generation AirPort Express shows a reduction of 41% (average) and 18% (peak) jitter on the digital optical output. Comparing jitter and THD on the analog output of the first and second generation AirPorts, it's clear there has been a major change between the units. Jitter and THD from the second generation AE has increased dramatically. This increase has the largest impact on over all performance, even for audiophiles using the digital output only. "(Emphasis and a little punctuation added.)

      So when looking only at jitter, you'd think the AE2 is the superior model. But when looking at the (more sonically important combination of) jitter and THD, the AE1 is the winner.
    1. HiFiGuy528's Avatar
      HiFiGuy528 -
      It's too deep for me to consume.
    1. JR_Audio's Avatar
      JR_Audio -
      To make my explanation a bit more clearer, I made a FFT of the FM Modulated Bi-Phase Signal (measuring the audio band related part of the jitter on the bi-phase digital output, similar what Ed Meitner has done in the past with his LIM detector on the master clock). Here you can see the impact of jitter into each audio frequency and you see, that the Airport Express 02 has in real, (real means what is left in the audio band), especially in the bass region, a much higher jitter than the latest version of the First Generation.

      Forget all the numbers that are given about the total amount, or peak peak, or average jitter of the digital out signal. This is only one aspect, but not the most important aspect. The most important question is, what will be the impact onto the audio signal, that we are listening to.

      Airport Express First Generation (latest production)



      Airport Express Second Genertation (actuall production)



      Juergen
    1. hearingaide's Avatar
      hearingaide -
      I don't know if the charts were competently created or not, but assuming they were, the jitter results from either generation would likely be inaudible. The noise floor of even awesome systems is usually well above the jitter dB levels shown in the charts. It would be interesting to see a valid double blind ABX test between the two generations with actual music to see the overall difference of each generation compared to the other. I also wonder if there are any gain difference between the two generations, which would make volume leveling for a valid comparison very important.
    1. roscoeiii's Avatar
      roscoeiii -
      Chris,

      I believe that there were more generations of the Airport Express before the one released in 6/2012. I think that the 2004 802.11g ones that Sterophile measured were replaced by a model often referred to as Gen 2 that came out in 2008, which upgraded to 802.11n.

      Details here:
      AirPort Express - Wikipedia, the free encyclopedia

      But regardless, thanks so much for these measurements.
    1. The Computer Audiophile's Avatar
      The Computer Audiophile -
      Hi roscoeiii - Thanks for the info. Maybe I should have called it gen 1 and gen 2 of the 802.11n version :~)

      From Wikipedia:

      • June 2004: AirPort Express released
      • March 2008: AirPort Express 802.11n (1st Generation) released
      • June 2012: AirPort Express 802.11n (2nd Generation) released
    1. hearingaide's Avatar
      hearingaide -
      As the writer notes, jitter is just one piece of the puzzle in audio quality, but comparing the chart below to the two charts in a few posts before this, I suspect the jitter is inaudible in both generations:

    1. myco's Avatar
      myco -
      Nice work on the measurements, Chris. Very interesting... Juergen, also nice amplification of the above. - Rob
    1. JR_Audio's Avatar
      JR_Audio -
      I respect the work of Julian Dunn a lot. I have lot of my knowledge about jitter from his work. I have known him from his early works at Nonophon and then with Prism Sound and finally with Audio Precision. His book „Measurement Technique for Digital Audio“ is one of my standard books, when it comes to understand all differnt jitter terms.

      But when his graph about the threshold of hearing jitter would be right, for experienced listeners, then we could stop all discussions and measurements about jitter and can stop looking for better oscizaltors, .... For me, the new Airport Express Second Generation is, via digital out, by far too bad, even for background listening.

      I do not want to open a discussion, what amound of jitter is audible and what amount not, but if this threshold of hearing jitter graph would be right, then every digital transport would sound the same and every software player would sound the same. Just take the measurend graphs for your information. What you are doing with this, is up to you.

      Juergen
    1. designmule's Avatar
      designmule -
      Quote Originally Posted by roscoeiii View Post
      I believe that there were more generations of the Airport Express before the one released in 6/2012. I think that the 2004 802.11g ones that Sterophile measured were replaced by a model often referred to as Gen 2 that came out in 2008, which upgraded to 802.11n.
      I have both a wireless G model & a wireless N model. My DacMagic frequently lost lock when used with the N but is rock solid with the G. At the time I was struggling with this issue, forums and such indicated that the jitter in the N model was much higher than the G. I recall at least one DAC manufacturer stating that they were trying to work with Apple to get it under control. FWIW, my N model is now connected to an Onkyo receiver and it has had no trouble with losing lock.
    1. mayhem13's Avatar
      mayhem13 -
      It was mentioned earlier the possible 'need' for a HiFi grade AirPlay compatible network audio player. I've often thought about this myself as a self contained network receiver, much like the currecnt Airport, but including a well implemented DAC with analog outs as well as digital pass through. AirPlay/AirTunes protocol has been 'cracked' so to speak and google results would lead you to the info. I would suspect that for a manufactured device, a license would be required from Apple at a substantial cost. Don't see why an open source project would be a trade infringement though. Food for thought if there's any developers out there. XBMC started out as a small DIY project on a forum and wound up being a huge part of the streaming/HTPC movement with commercial developers moving towards dedicated ARM processors to run it natively. Might be a great DIY/CA project that the whole audio streaming community could enjoy. Surprised one of the DAC mfgrs hasn't moved in the direction of an AirPlay/DNLA capable network DAC. Cambridge has a network player, but it's not AirPlay enabled. Denon, Pioneer and Marantz have AirPlay enabled players but I wonder if they focus enough on DAC Implementation or more on functionality?
    1. ringenesherre's Avatar
      ringenesherre -
      Quote Originally Posted by JR_Audio View Post
      I do not want to open a discussion, what amound of jitter is audible and what amount not, but if this threshold of hearing jitter graph would be right, then every digital transport would sound the same and every software player would sound the same.
      That sounds like an entirely plausible conclusion. Of course, given that the transport is reasonably well engineered (as the Airport Express seems to be).

      Cheers,
      Peter
    1. 1audio's Avatar
      1audio -
      Jurgen:
      Please post more details on your settings for the jitter measurements so others can duplicate them. I believe they are standard for the AP but may not match other systems.

      Jitter is one of many aspects that can impact the audio quality of an audio chain. While it may be below the threshold of audibility the chart published is old work and newer research may show different levels. However, getting it right is no harder than getting it wrong in this case. It requires some effort at the original design stage but these levels do not require heroic levels of engineering and expensive components. They do require attention at the design and acceptance stages, something Apple is usually very good about. I don't believe Apple devotes much attention to audio quality as we define it. It has not hurt their bottom line.
    1. JR_Audio's Avatar
      JR_Audio -
      FM Demodulation

      Basically, this is a FM Demodulation and analysing the result with an FFT, calibratet do time.

      Ed Meitner was the first (to my knowledge), doing something similar. He was looking at the master clock, direct at the master clock pin of DA chips. Here you have to know what oversampling rate the unit is running, then you tune a fixed, low jitter oszilator to that master clock frequency, and use a FM Tuner as demodulator to get the FM part (= Jitter) out of the master clock.

      When Julian Dunn was working for / with Audio Precision (his above mentioned book is a very, very good source about Jitter and more), they devoloped a mention method even one step further, to FM demodulate the Bi-Phase Signal. The exact method for measureing the Bi-Phase signal is maybe a step too far at this stage, but …

      You can search for some old literature about Ed Meitners LIM detector and also the highly recommended book from Julian Dunn (even when is limits about hearing of jitter is, at least in my opinion, by far too high), but he has a very clear writing about jitter. You will find also some of his earlier work with Nanophone on the web. Good luck.

      Juergen